Research On SLAM Technology Of Mobile Robot Based On Multi-Line Lidar

With the development of technology,robots are slowly entering the public's field of vision,and mobile robots as an important part of the robot family,are also developing rapidly.Among them,simultaneous localization and mapping in an unknown environment are necessary conditions for the mobile robot's subsequent navigation and environment perception and key technology in the process of mobile robots finally moving towards full intelligent movement.In this paper,the multi-line lidar is used as the main sensor for the outdoor plant environment without GPS,and the inertial measurement unit is used as the auxiliary sensor to study the problems of simultaneous localization and mapping of the logistics mobile robot in the plant area.The main research content of this paper includes four parts: multi-line lidar raw data preprocessing,inter-frame motion estimation,environment map construction and positioning,algorithm comparison and simulation verification.In view of the problem of motion distortion and interference points in the point cloud data collected by multi-line lidar,the causes of motion distortion and interference points are analyzed.The IMU information is used to correct the motion distortion of the laser point cloud and a method based on depth map is used to effectively eliminate interference points.In order to improve the accuracy of the relative motion between adjacent frames in the laser SLAM system,this paper improves the traditional point cloud matching method for obtaining relative motion,and uses the method of multi-line lidar and IMU data fusion to obtain the relative motion between adjacent frames.By verifying the comparison of data sets,this method can solve the relative motion between adjacent frames more accurately.Without the absolute position provided by GPS,the positioning error of the robot will gradually increase with time.In order to solve this problem,this paper uses a loop detection method based on joint judgment of position constraints and ICP constraints,and uses the ISAM algorithm to modify the detected pose of the loop,effectively reducing the positioning error of the mobile robot.In addition,the way to update the map is improved,and the map is decomposed into a form of a set of key frame poses and a point cloud set corresponding to the key frame.After the loop correction is completed,the map is reconstructed,which effectively reduces the appearance of map ghosting.The sequences in the KITTI data set are used to compare the relative trajectory error,absolute trajectory error,and mapping effect of this method and the LOAM method,and it is confirmed that this method is superior to LOAM.At the same time,in the Gazebo simulation environment,the outdoor plant environment and the outdoor logistics robot are simulated,and the feature point extraction,positioning error analysis,mapping effect and algorithm time-consuming analysis are analyzed in this environment.The results show that the method of this paper can work effectively in this environment.In short,this paper improves the positioning accuracy and mapping effect of the traditional laser SLAM system,and the method in this paper can meet the corresponding index.